The advance of digital image processing techniques has allowed image forming apparatuses such as copying machines and printers based on an electrophotography method or an ink-jet method to reproduce color images with high quality, and accordingly digital copying machines and digital multifunction printing apparatuses with full colors have been commercially available.
Documents to be copied by such image forming apparatuses include text areas, halftone dot areas, line areas, and continuous tone image areas (photographic-pictures). Further, these areas may be mixed in a single document. As such, in order to obtain a good reproduction image (copied image), it is necessary to perform the most suitable image process according to the type of areas included in each document.
Therefore, a so-called segmentation process is performed in such image forming apparatuses. The segmentation process is a process in which: in image data to be processed, a mask centering a current pixel is used to find “an image area into which the current pixel is classified”, and a segmentation class signal indicative of the image area is outputted. Examples of the image areas classified by the segmentation process include text areas, halftone dot areas, and other areas (areas which are not classified as text areas and halftone dot areas, such as continuous tone image areas and page background areas).
Generally, in the segmentation process, an input image read by an image input apparatus such as a scanner is separated into blocks each constituted of a plurality of pixels and each pixel or each block is judged to be (identified with) one of a text area, a halftone dot area, and other area. Then, based on the result of the judgment, an image process most suitable for the area is performed according to the type of areas included in a document.
Further, as an example of the segmentation process, there is known the technique disclosed in Document 1: Japanese Unexamined Patent Publication No. 199588/1990 (Tokukaihei 2-199588; published on Aug. 7, 1990). The following shortly explains the procedure of the technique. In the technique disclosed in Document 1, in view of characteristics of a page background (wide area, uniformity and continuity of density (pixel value) or color), an area having the same density as the page background or an area having a uniform color is judged to be a page background density area. Further, it is judged whether the page background density area is a page background area or not in accordance with the size or the like of the page background density area. Thereafter, out of a non-page background area, an area whose size is not more than a predetermined size is identified as a text area and an area whose size is more than the predetermined size is identified as a halftone dot area.
Further, in the above image forming apparatus, it is necessary to perform not only the segmentation process but also correction of density of an input image. In terms of this point, Document 2: Japanese Unexamined Patent Publication No. 130398/1993 (Tokukaihei 5-130398; published on May 25, 1993) discloses a full color imaging apparatus having (i) a function for allowing a user to set γ characteristics (a relation between density of a document image and density of a copied image) which are recommended in advance so as to correspond to the type of a document, merely by selecting the type of the document, and (ii) a function for allowing a user to exactly adjust γ characteristics. The full color imaging apparatus allows exact adjustment of γ characteristics at a time when a person skilled in printing (such as designer or the like) performs copies, while the full color imaging apparatus allows easy setting of γ characteristics suitable for the type of the document merely by selecting the type of the document at a time when a normal user performs copies. Therefore, the technique allows a user to set γ characteristics with an operation function suitable for the user's skill.
Further, Document 3: Japanese Unexamined Patent Publication No. 171404/2002 (Tokukai 2002-171404; published on Jun. 14, 2002) discloses an image processing apparatus capable of performing a segmentation process and a density conversion process in this order.
In the above image forming apparatus, in order to enhance readability of a text in printed images, a text area is detected in the segmentation process and an edge enhancement process is performed with respect to the text area. As such, it is necessary to detect a text area with high accuracy in the segmentation process.
Generally, in the segmentation process, a high density part on a white page background area is detected as a text area on the premise that “a text normally exists on a white page background”. As such, when the segmentation process is performed with respect to a document image whose density is corrected so that page background density is removed, it is possible to detect, as text areas, not only a text surrounded by a white page background but also a text surrounded by a page background having density (such as a text on news papers), so that it is possible to detect a text with higher accuracy.
However, when the segmentation process is performed with respect to the image whose density is corrected so that page background density is removed, there is such a problem that an area other than a text is detected as a text area. For example, in a case where a photographic-picture area is included in an image, when the density correction is performed, density of a low density area surrounding an edge portion of the photographic-picture area may be removed. At that time, the edge portion is detected as a high density part existing on the white page background. As a result, the edge portion is wrongly detected as a text, and accordingly not only detection accuracy of a text area but also detection accuracy of the photographic-picture area drops. As such, it is impossible to perform the segmentation process with high accuracy.
On the other hand, in order to suppress wrong detection of the edge portion of the photographic-picture, the segmentation process should be performed based on the image whose density is not corrected. At that time, even when a low density area exists around the edge portion of the photographic-picture, the segmentation process is performed based on an image which keeps enough density in the low density area. As a result, the edge portion is not detected as the high density part on the white page background, so that it is possible to suppress wrong detection of the edge portion as a text. However, in a case where the segmentation process is performed with respect to the image which has not been subjected to the density correction, density of a page background having certain density is not removed, so that it is difficult to detect a text on the page background having certain density (a text on news papers). As a result, it is impossible to perform the segmentation process with high accuracy.
Namely, whether the density correction is performed or not before the segmentation process depends on the kinds and characteristics of a document image to be processed. In order to perform the segmentation process with high accuracy, the order of the segmentation process and the density correction process should not be fixed.